This is the pre-peer-reviewed version of the following article:

Transcription

1 This is the pre-peer-reviewed version of the following article: Legislator Success in Committee: Gatekeeping Authority and the Loss of Majority Control Ernesto Calvo University of Houston, Houston University of Maryland, College Park Iñaki Sagarzazu University of Houston, Houston Nuffield College, Oxford American Journal of Political Science April/2010 Abstract: In multi-party legislatures, the largest party or coalition may fall short of controlling a majority of plenary seats. However, plurality-led congresses generally endow the largest parties with extensive agenda setting prerogatives, even when plenary majorities are lacking. In this article we present a model and compelling evidence describing changes in the strategic behavior of committee chairs when majority control is lost. Using a dataset that includes all the bills proposed to the Argentine House in the last twenty-five years, we estimate success in committee in majority- and plurality-led congresses. We provide extensive evidence that the loss of majority control reduces the importance of the median voter of the plurality party while improving the success of the median committee voter. * Acknowledgments: We thank for their suggestions Juan Manuel Abal Medina (jr), Manuel Alcántara, Isabella Alcañiz, Eduardo Aleman, Maria Baron, Miguel de Luca, Marcelo Escolar, Tulia Falleti, Flavia Freidenberg, Mariana Gutierrez, Eugenio Inchausti, Noah Kaplan, Juan Pablo Micozzi, Mercedes García Montero, Maria Victoria Murillo, Ana Maria Mustapic, Fernando Pedrosa, Susan Scarrow, Gisela Sin, Andres Tow, and the participants of the Latin America Workshop at the University of Houston. Keywords: Congress, Standing Committees, Legislative Success, Agenda Setting.

2 In multi-party legislatures, the largest party or coalition may fall short of controlling a majority of plenary seats. 1 However, congressional rules generally endow the largest party with extensive agenda setting prerogatives, even if plenary majorities are lacking. How do plurality parties adjust to such partisan environments? Is the result more pronounced legislative gridlock or a change in the makeup of the legislation to be considered in committee and voted on the plenary floor? In this article we analyze the determinants of legislative success in committee 2 and provide compelling evidence that the loss of majority support significantly alters the agenda setting strategies of committee chairs. We show that, in plurality-led congresses, committee chairs pursue more permissive reporting strategies and delegate further gatekeeping responsibilities onto the Chamber Directorate. To account for legislators' success in committee, we extend Cox and McCubbins' (2005) model of a dual-veto system to plurality-led congresses. We show that resources available to senior party members to prevent legislation from being reported from committee far exceed their capacity to sustain quorum on the plenary floor or to muster the votes required for final passage. 1 Data, code, and ancillary materials to replicate all analysis may be downloaded from 2 We use the term success in committee to describe success in discharging a bill for consideration for further consideration by the plenary floor. In the US Congress literature, Anderson et.al. (2003) provide the only systematic analysis we know of success in committee. On a related research, Krutz (2005) presents a novel analysis of winnowing, measuring the probability that a bill will be placed for the consideration of Committee members (Krutz 2005). Our focus on the success of individual legislators differs from prior research explaining how the loss of majority control affects executive success (Figuereido, Cheibub, and Limongi 2000; Mayhew 1974; Howell et al. 2000; Cheibub, Przeworski, and Saiegh 2004; Canes-Wrone and Marchi 2002; Saiegh 2009) or how fragmented Congresses limit (or not) the range of policies available to the President (Amorim Neto, Cox, and McCubbins 2003; Ames 2001). 1

3 Consequently, committee chairs with extensive scheduling prerogatives need to bargain with minority leaders, who may withhold legislation through the enforcement of quorum and reporting rules. To explain legislator success in majority- and plurality-led congresses, we analyze in detail twenty-five years of lawmaking in the Argentine House of representatives. We study legislator success in committee using over 29,000 law initiatives proposed by members of the Argentine House. Results show that the loss of committee majorities has little effect on the success of plurality-party members but results in a significant increase in legislative success among members of other parties. Overall, the loss of majority support increases the total number of bills reported from committee, which is accompanied by a small ideological drift towards the median voter of the chamber. The importance of understanding the effect that the loss of majority control has on legislative success goes beyond the particular case of Argentina. In a study of legislative politics in Latin America, García Montero (2009) notes that the largest House coalition failed to gather a majority of seats in 33 out 70 legislative periods or 47% of her sample. Moreover, the largest party felt short of controlling a majority of seats in 48 out of those same 70 periods, a very significant 69% of the sample. Considering all presidential democracies in the world since 1950, the largest party in the lower chamber failed to obtain a majority of seats in 43% of congressional periods (Cheibub, Przeworski, and Saiegh 2004). Thus far, however, limited research efforts have been directed towards understanding legislators' success in plurality-led congresses and no research that we are aware of explains success in committee. 3 3 There is, however, a significant literature that analyzes legislative success of minority presidents (Alemán and Calvo 2010; Ames 2001; Amorim Neto, Cox, and McCubbins 2003; Cheibub, Przeworski, and Saiegh 2

4 As in the US Congress, in most Latin American legislatures a majority of the bills die at the committee stage. As shown in Aleman (2006), approximately 50% percent of legislatures in Latin America impose no deadlines for the consideration or reporting of legislation to the plenary floor. Although an emerging literature has taken notice of the importance of committees in the organization of Latin American legislatures (Pereira and Mueller 2004; Santos and Renno 2004; Crisp et al. 2009; Aparicio and Langston 2009), no research that we are aware of has explored the determinants of success in committee. While little research analyzes the working of committees in the presidential regimes of Latin America, an important body of research exists which details the workings of the committee system in the US Congress (Cox and McCubbins 1993, 2005; Krehbiel 1996; Weingast 1989). The conventional story, masterfully presented by Gary Cox and Matthew McCubbins in Setting the Agenda, presumes that rational anticipation by committee chairs serves the purpose of screening legislation that may divide and defeat the majority party on a floor vote. However, in plurality-led congresses -which are never observed in the US- rational anticipation is not enough to overcome reporting and quorum rules that grant minority leaders considerable bargaining leverage. As we will show, changes in the composition of committees and the threat of a vanishing quorum result in a significant change in the type of legislation reported from committee and approved on the floor. The organization of this article is the following: in the next two sections we discuss how the loss of majority control affects success in committee. We describe the gatekeeping prerogatives of committee chairs and provide a spatial model to describe the voting environments in majority- and plurality-led Congresses. In the fourth section we describe the data, variables, and modeling strategy used to analyze success in reporting bills from committee. In the fifth section we discuss 2004; Figuereido, Cheibub, and Limongi 2000; García Montero 2009; Pereira and Mueller 2004; Saiegh 2009). 3

5 statistical results explaining the determinants of success in committees. We conclude in the sixth section. Committee Success with Decentralized Gatekeeping Authority In the last few years, significant research in Latin America has been conducted that sheds new light on the organization of congresses and the behavior of legislators (Amorim Neto, Cox, and McCubbins 2003; Carey 2007; Desposato 2006; Morgenstern 2004; Crisp et al. 2004; García Montero 2009; Jones and Hwang 2005; Taylor-Robinson and Diaz 1999). Data collection efforts and new statistical techniques now allow researchers to analyze the ideological distribution of parties and factions in Latin American legislatures and to understand better their internal organization. A broad scholarship has emerged, shedding new light on a range of important problems such as the determinants of party discipline (Carey 2007; Desposato 2006), the stability of presidential coalitions (Aleman and Tsebelis 2005; Figuereido, Cheibub, and Limongi 2000), the organization of authority roles, and the determinants of legislative success (Amorim Neto, Cox, and McCubbins 2003; Crisp, Kanthak, and Leijonhufvud 2008; Figuereido, Cheibub, and Limongi 2000). In this article we further contribute to this literature, analyzing the relationship between gatekeeping authority, majority control, and legislative success. To explain the relationship between gate keeping authority and legislative success, we focus on the mechanisms used by parties to regulate the flow of legislation reported to the floor (Shepsle & Weingast, 1987; Cox and McCubbins, 2005). In most Latin American countries, political parties play a crucial role organizing the daily business of the legislature. These parties are not only the key players controlling the deck of legislation reported to the plenary, but they are also endowed with a variety of resources to restrict the set of policy choices available to legislators and to affect their voting behavior (Amorim Neto, Cox, and McCubbins 2003). An important mission of legislators is 4

6 the enactment of laws they favor and the rejection of those they dislike. But in legislatures where power is not distributed equally, not all parties will be equally successful at advancing their members lawmaking goals (Cox 2006). In a party-based legislature, it is to be expected that bills spearheaded by members of the majority party will have a higher probability of passage than those proposed by members of other parties. In the time-constrained environment common to all legislatures, the party leaders with actual control over the scheduling of proposals should be expected to prioritize the bills of their fellow partisans. Partisan biases in the time allocated to each initiative, however, compound as bills move forward through successive legislative stages. Agenda setting prerogatives that restrict the deck of bills to be considered by committee members, such as wider discretion by committee chairs to withhold the discussion of a particular proposal, will result in a more heavily partisan sample of bills reported to the floor (Cox and McCubbins 2005). 4 Wider agenda setting prerogatives by the Chamber Directorate will grant the majority party with resources to restrict further this already partisan sample of bills. In this second stage, agenda setting prerogatives will more clearly serve the purpose of advancing legislation that is both important to the senior partners of the majority party and that will not divide its members when reviewed by the plenary. In the Argentine case, the sequential organization of the legislative process also biases the sample of bills receiving final approval towards the majority party. In the earlier stages, gatekeeping authority is decentralized to senior party members who control important committee posts, 4 Cox and McCubbins (2005) extensively describe, in chapter 6 of Setting the Agenda, the heavily partisan nature of the bills reported to the floor: For the Democratically controlled 82 nd Congress, 26% were sponsored by minority party members, while it was 13% in the Republican- controlled 83rd Congress. Thus, the Democrats sponsorship rate fell from 74% to 13% when their seat share fell from 53.8% to 49.0% (p. 268). 5

7 endowing the majority party with institutional resources to screen unwanted legislation in their respective jurisdictions, e.g. the appropriations, municipal affairs, constitutional affairs committees, etc. Decentralized gatekeeping authority at the committee level, however, may results in divisive legislation that is not unanimously supported by all senior partners of the party squeezing their way to the floor. 5 Consequently, a second safety valve is in place to filter bills that sneak out because of the decentralized nature of the committee review process. Legislation that is reported from committee is subject to review in pre-floor party meetings, before they can be scheduled for a plenary vote by the Chamber Directorate (Comisión de Labor Parlamentaria). A second distinctive feature of the Argentine Congress is that Committee chairmanships are allocated to parties in proportion to their share of Chamber seats. Consequently, a significant number of committee chairs belong to minority parties. While the majority party always controls the most important committees, such as the appropriations committee (Presupuesto y Hacienda), opposition chairs dispose of institutional resources to advance some of their preferred policies in committee. This feature reinforces the importance of gatekeeping authority in later stages and a more active bargaining strategy than in the US Congress. While committee chairs are proportionally assigned in accordance to house seats, the composition of committees replicates the overall partisan distribution of seats in the House. Consequently, the majority/plurality party controls a majority/plurality of seats in almost all committees. 6 5 This may occur because committees have preferences that differ from the floor or because the pay attention to issues on which no senior party members has previously expressed a preference. For an analysis of issue attention in committee see Sheingate (2006). 6 In the last twenty five years, there have been only eight instances (<2%) in which an opposition party held the majority or plurality of seats in a committee. All eight cases occurred in politically marginal committees. 6

8 Because of the importance of late review institutions, all amendments to bills reported from committee (modificaciones) are attached to a unified dossier which includes the majority report, minority reports from dissenting members of the committee, and observations or objections raised by fellow representatives. Each proposed amendment must be made available to members of Congress a week before the bill is schedule for plenary consideration. 7 The chamber directorate, however, has considerable discretion when deciding which bills should be placed on the plenary agenda. Consequently, as it will be shown later, close to half of the initiatives reported from committee will never reach the floor. Finally, special majorities are required to propose further amendments once the bill reaches the plenary, making the vote a horse race between proposals introduced ahead of any floor debate. The makeup of the floor agenda is, therefore, sequentially restricted to represent better the preferences of the majority party members, at the committee level, and screened for potential conflicts of interest in later stages. This process is then repeated in the alternate Chamber. The result of this dual screening process is the elimination of unwanted legislation proposed by minority parties in the early stages, and the promotion of legislation that is unanimously preferred by the senior members of the majority party in the later stages. 8 7 Restrictions to amendments introduced on the floor also deter the passage of legislation through irregular scheduling mechanisms. The expression entre gallos y medialunas (night time sessions), for example, describes the strategy of voting legislation that is brought to the floor late in a debate or in special sessions with a minimum quorum. Last minute amendments still occur when extraordinary circumstances call for it, generally on very controversial initiatives requiring intense bargaining within and across legislative blocs (Calvo and Tow 2008). 8 These pre-floor party meetings are responsible for screening out 40% of the legislation reported from committee. 7

9 On the Discharge of Legislation from Committee and the Loss of Majority Control As described before, the sequential organization of the legislative process provides the majority party with significant resources to (i) screen out unwanted legislation at the committee level, (ii) screen out divisive legislation in the pre-floor party meeting, and (iii) screen out bills that lack majority support in the Chamber Directorate meeting. These three screening stages all but guarantee that a bill that is submitted to the plenary will receive a favorable vote: of the 29,173 bill initiatives submitted to Congress by individual deputies between 1984 and 2007, only 5 were rejected in a floor vote! 9 Meanwhile, over 1,665 bills received a favorable vote in the floor (5.71%) and 179 projects were withdrawn from consideration. The loss of majority control, however, substantively affects this process. Below, we provide a streamlined model describing how the loss of majority control affects legislative success in committee. Success in Committee: a formal description Different from final passage in a plenary vote, success in committee is achieved if and when a bill is discharged with a majority report. Such a report requires the signature of a majority of committee members. Bills without a majority report cannot be debated on the plenary even if a minority report is drafted. Consequently, the loss of majority control in committee will prevent the plurality party (the party with most seats but short of a majority), from automatically forcing a bill 9 Another 77 bills (0.26%) changed their status to private bills. It is worth noting that reaching the floor does not guarantee approval. However, bills that reach the plenary and lack support tend to be reported back to committee for further consideration, rather than voted down on the floor. 8

10 onto the plenary floor. If a majority report is lacking, discharge motions can be introduced on the plenary by individual deputies, requiring the support of 2 3 of plenary attendees. The high threshold required to force a bill onto the floor without a majority report makes it very difficult to circumvent the committee gates. Because a simple majority is required to discharge a bill from Committee, the loss of majority control grants the median committee member, who may now be a member of the opposition, a decisive vote to defeat proposals submitted by the plurality party. 10 As we will show, the Committee and plenary process in Argentina replicates the dual veto system of the Pre-Reed US House, where the opposition may prevent the reconsideration of policies disliked by its own median voter. For presentation purposes, we begin by revisiting Cox and McCubbins (2005) exposition of the dual-veto system in the US House and then describe how the agenda setting prerogatives of chairs affect legislative success in committee. 11 Finally, we provide a general model to explain legislative success in plurality-led congresses and describe testable implications. As with most positive models of Legislative activity, we begin with a set of simplifying assumptions where k committee members consider a single proposal,. Each member is characterized by separable, single peaked preferences, described by unique ideal points in a 10 The median voter will only be a member of an opposition party for extreme plurality parties. This is unproblematic assumption in the Argentine Congress, where ideal point estimates consistently placed the Peronists (PJ) and the Unión Cívica Radical (UCR) on the extremes of the distribution. 11 In discussing vote decisions by committees, Cox and McCubbins (2002) notice that the conditional cartel model can be easily translated to committee environments. They coin the term local agenda control when distinguishing success in committee from the global agenda control in the plenary floor. 9

11 single dimension j. We also assume a quadratic loss function 12 measuring Euclidian distances between a proposed policy and the legislator s preference,. Each legislator belongs to a single legislative bloc, with committee membership determined by House rules. To simplify the presentation of the model, we consider Deputies as members of one among three available legislative blocs: a majority party whose median voter is described by ; a first minority party whose median voter is described by ; and a second minority party whose median voter is described by. It is important to note that with three or more blocs, a legislative party controlling most seats could still fail to have as member the median voter of the Committee. Consistent with the description of the dual veto system in Cox and McCubbins, we locate the median voter of the majority bloc to the left of the first minority median voter, which is also to the left of the second minority median voter,. This depiction adjusts well to the Argentine House, characterized by spatially extreme majority/plurality parties (Aleman et al. 2009; Jones and Hwang 2005). Consequently, in our majority-led Congress, the median committee voter C lays to the right of the median voter of the majority bloc and to the left of the median voter of the first minority ;. In accordance with Committee rules in the Argentine House, we consider the Chairman of each Committee as the sole agenda setter, endowed with exclusive scheduling authority. 13 If a bill is open to consideration, committee members vote sincerely and maximize their preferred outcome, e.g. voting YEA if the utility of the proposed 12 The choice of a quadratic loss function instead of Cox and McCubbins absolute deviation, :, will become more important in the statistical analyses of sections In Argentina the drafting of the committee schedule is the exclusive responsibility of each committee chair. Chairs may also request feasibility studies on a particular bill and their vote counts double for reporting decisions 10

12 policy is higher than that of the current status quo,, or voting NAY if status quo is preferred to the proposal,. In a first description of our committee model, amendments can be proposed during debate (open rule), guaranteeing that the final policy result will reflect the preferences of the median committee member, C. << Insert Figure 1>> Figure 1.A provides a single dimension representation of a majority-led committee with a chair drawn from the majority bloc. The area between the median committee voter and 2 describes the blockout zone, where the reconsideration of the status quo in a policy jurisdiction would divide the majority bloc and move policy further away from the preferences of the median member of the majority party, e.g. moving policy from to. Consequently, as depicted in Figure 1.A, committee chairs from the majority party should prevent the reconsideration of, which is located to the left of the median committee member. By contrast, the chair should allow the reconsideration of, redirecting policy towards the majority party, as described by the arrow connecting and C. As described in Figure 1.B, however, the loss of majority control will allow senior members of the minority party to prevent the reconsideration of policies that divide their own legislative bloc and move policy away from their own median voter. As in the dual-veto system of the pre- Reed house (Cox and McCubbins, 2005), the loss of majority control provides senior members of minority parties with the capacity to deny committee (or plenary) quorum when the proposed policy leads to the reconsideration of a status quo in the area between the median committee member and 2, where. House rules that bestow committee chairs with the capacity to restrict the set of amendments to be considered, allow the majority party to propose policy changes that move the status quo further closer to the preferences of the median voter of their party. Relaxing the assumption of open 11

13 rule, as in Figure 1.C, allows the median committee member of the majority party,, to propose policy which is preferred to by the median committee voter C. Finally, when majority control is lost, the minority party should prevent the reconsideration of any policy to the left of but allow limited reconsideration of a status quo to the right of if it is preferred by its median bloc member. As described by Figure 1.D, the first minority party should facilitate quorum for both the reconsideration of and. However, the acceptable range of policy change where is considerable narrower than for. Two important results should be highlighted to understand the effect that the loss of majority control will have on success in committee: (i) First, a majority-led committee with restrictive proposal authority will endow committee members of the majority party with broader discretion to select proposals that move policy beyond the median committee member and closer to the median member of the majority party. (ii) Second, the loss of majority control will restrict policy change to proposals that are acceptable to the median voter of the majority party and will not be vetoed by the median member of the minority party. Because membership in committee is reflective of overall party shares, the loss of majority control in the Chamber will also result in a loss of majority control in most committees. 14 Significant institutional authority by the committee chair, however, will still provide the plurality party with plenty of instruments to restrict legislation that is reported to the floor. Should the newly acquired veto authority in plurality led committees result in a decline in the amount of legislation reported from committee or a change in the ideological make-up of the plenary agenda? As we will show in the next section, the gatekeeping strategy of committee chairs will vary, conditional on the context on the plenary floor. When a party controls a majority of seats 14 Committees in Argentina have a larger number of members than in the US Congress and membership shares roughly reflect the proportion of total Chamber seats held by each party. 12

14 on the plenary floor, chairs will favor the discharge of legislation preferred by their median voter even after losing majority support in Committee, which will then result in a small decline in the total amount of legislation reported to the floor. By contrast, in plurality led congresses, committee chairs will become more permissive, allowing a larger set of bills to be discharged and delegating further gatekeeping authority to pre-floor party meetings and the chamber directorate. The result, as we will show, is a higher rate of success in committee and a change in the ideological make-up of the plenary agenda. << INSERT TABLE 1>> Table 1 presents preliminary evidence that strongly conforms to the proposed model, describing the percent of bills reported to the floor by the two largest legislative blocs in Argentina, the Peronist Justicialista Party (PJ) and the centrist UCR. When Peronists control a majority of House seats, the success in reporting bills sponsored by members of the party is 16%. The loss of majority control results in a very small decline in the Peronists sponsored legislation discharged from committee and a more substantial increase in the success of legislation sponsored by other blocs. The increase in the share of bills reported by other parties significantly alters the makeup of the total bills reported from Committee. Notice that, while the Peronists success in committee remains almost unchanged -a decline of 0.3%-, the share of Peronists bills is 12% smaller. We observe very similar results when the UCR loses majority control, with a mild decline in the success of UCR legislators paired with a significant increase in the legislation reported by members of other parties. Table 1 already indicates that the loss of majority control results in a change in the partisan composition of the bills reported to the floor, rather than a systematic decline of success in committee. In fact, the overall percent of bills reported by all committees tends to be larger in plurality controlled congresses, with gatekeeping being delegated to the party pre-floor meeting and 13

15 the Chamber Directorate. In the next section, we use a very large dataset of bill initiatives proposed to the Argentine Congress since 1984 to explain the determinants of Committee success under majority and plurality controlled Congresses. Explaining Committee Success in the Argentine Congress To analyze how the institutional organization of Congress determines success in committee, we use a dataset of 29,173 legislative initiatives, which include all public bills proposed by Deputies to the Argentine Congress from 1984 to Recent advances in the measurement of ideological preferences of legislators (Aleman et al. 2009) allow us to use co-sponsorship data for all available bills to retrieve ideal point estimates describing the spatial preferences of House members. It is important to highlight that the activity of co-sponsoring legislation, which we use to draw ideal point estimates of the legislator s preferences, takes place before a bill is considered in committee or discussed on the plenary floor. Consequently, revealed preferences about the legislators preferences in co-sponsorship data are not affected by Committee and Plenary behavior and will not be endogenously related to our variables of legislative success. Our strategy for modeling success in committee is to run a multilevel logistic regression to estimate the probability that a bill will be successfully reported from committee in response to contextual changes in the committee and on the plenary floor. The choice of a multilevel model is appropriate, as we have information that is not only specific to each bill, but also information that is specific to each legislator (such as the level of seniority or party membership), contextual 15 We restrict our analysis to the Lower Chamber because the Peronists never lost majority control in the Argentine Senate. It is possible, however, that rational anticipation lead to the introduction of amendments in later legislative stages. Sin and Lupia produce a model interpreting the effect in the House of a change in the partisan composition of the Senate or the Presidency (Sin and Lupia 2008). 14

16 information available at the committee level (such as whether the largest party in the committee has a majority or plurality of seats), and information at the Congress level. A multilevel design allows us to control for the proper levels at which each type of data is collected. Because the Argentine Congress provides a fixed schedule for the consideration of bills in each Chamber, there is no selection bias when estimating success in committee. In Argentina, all bill initiatives are first evaluated by House personnel, who decide committee referrals and the number of readings in a non-partisan way. 16 After consideration and discussion, successful initiatives are discharged with accompanying reports and suggested amendments. A specialized agency, the Secretary of Parliamentary Affairs, distributes the proposed bill, amendments, and ancillary materials to all representatives at least a week before it can be scheduled for a floor vote. Any representative can propose further amendments prior to the scheduling of the bill, but special super-majority requirements are in effect during the plenary debate. Regularly scheduled party meetings then discuss these proposals and the Chamber Directorate makes the final decision to place a bill on the plenary agenda. Finally, bills that are successfully scheduled for plenary consideration and receive a positive vote are shuttled to the alternate Chamber. This fixed schedule guarantees that a smaller sample of bills moves to each successive stage. In Table 2 we provide descriptive information about success in each of these stages. 16 Each committee represents a reading of the bill initiative with the lead committee as the first reading. While Congressional rules require for every project to receive a joint signed report after an inter-committee meeting, it is customary that committees sign the joint report sequentially. The reason is that a joint meeting of three or four committees would include between 60 and 80 representatives, making it virtually impossible to meet the formal criteria of a joint meeting. Since 2007, the number of committees that read a project has been summarily restricted to 2. However, a dictamen to report the bill to the floor requires the signature of all committees that participate from the reading (lectura). 15

17 << INSERT TABLE 2 >> As shown in Table 2, changes in House support result in a different makeup of the legislation that receives dictamen at the committee level and reaches the floor. In both the Peronists and UCR controlled houses, the party controlling at least a plurality of the seats is able to report a larger share of bills initiated by its members, 13%. Other third parties, however, are still able to report to the floor a respectable 8% of their preferred legislation. 17 Measuring Success in Committee We run a number of models estimating the probability that a bill will be successfully reported from committee. This variable, success in committee, takes the value of 1 if the proposal receives a joint dictamen (dossier reporting the bill for further consideration by the Chamber) or the value of 0 if it dies in committee (cajoneada). Irrespective of how many committees participate from the discussion, bills need to be discharged from all committees with jurisdiction on the proposed policy. Because bills may be referred to multiple committees, the 29,173 bill initiatives provide us with 61,892 observations. We also estimate a separate model measuring success on the plenary floor, with a dependent variable taking the value of 1 if approved in the House and 0 otherwise. This alternate model only includes bills successfully discharged from committee and will allow us to provide further evidence of legislative success after the committee stage. 17 Majority control, consequently, has a more moderate effect on committee reporting than is commonly observed in the US Congress. As shown by Cox and McCubbins (2005), close to 80% of bills reported from committee in the US are sponsored by members of the majority party. The difference steams from the control that minority parties have of a proportional number of committee chairmanships even in majority controlled congresses. 16

18 To explain success in committee, we consider various behavioral, institutional, and contextual variables at the bill, legislator, committee, and plenary level. Most of our interest, however, centers on how committee success is affected by the ideological proximity between the sponsor of the bill and the median legislator of the majority coalition. Ideological determinants of Legislative Success We first turn our attention to the ideological determinants of legislative success. We use two different independent variables measuring the (i) squared ideological distance between the lead sponsor of each bill and the median committee member; and (ii) the squared distance between the lead sponsor of the bill and the median committee member of the majority party. 18 Using these two measures of ideological proximity we are able to map success in committee for any region of the ideological space. The ideological location of legislators was retrieved using principal component analysis on the agreement matrix of cosponsored legislation (Aleman et al. 2009). This procedure is a family relative of Keith Poole s Optimal Classification design (Poole 2005), retrieving ideal points estimates from cosponsorship data. An advantage of our choice of ideal points is that estimates are not retrieved from nominal votes but, instead, from revealed preferences that occur prior to any legislative activity in committee. In the Argentine Congress, a lead sponsor (firmante) requests fellow legislators to co-sponsor a bill before it is formally proposed to the Chamber. Consequently, 18 Different from the US Congress, in Argentina bills have an official lead sponsor (firmante) and multiple cosponsors (cofirmantes). Over 61% of bills are signed by the lead sponsor alone and 90% have fewer than four cosponsors. Consequently, the ideological location of the sponsor of the bill is very informative about the overall policy location of the proposal. For detail information about cosponsorship in Argentina see Aleman et.al. (2009). 17

19 estimates drawn from cosponsorhsip data take precedence and are independent from observations of success in committee or plenary consideration. 19 Our expectation is that the probability that a bill will be reported from the committee to the floor will increase when the sponsor (firmante) is closer to the median committee member of the majority party. This relationship, however, should lose steam when a party loses majority support on the plenary floor and when committee chairs are members of an opposition party with control over the committee schedule. In those cases, committees should become more permissive, delegating gatekeeping responsibilities to pre-floor party meetings and the chamber directorate. Figure 1.A in the appendix provides a visual representation of the distribution of ideological preferences of representatives (first dimension) in twelve Argentine Congresses. It also provides descriptive information about the location of the median voter on the plenary floor and the location of the median voters in each of the standing committees of the House. The Context inside Committees As indicated in section 3, we expect that the context in committee will endow the minority party with resources to prevent the reconsideration of legislation that is not preferred by its median voter. These contextual features are captured by the two main independent ideological distance variables described before. Because we are already estimating separate models for majority/plurality-led congresses and for majority/minority chairs, there is little contextual variation to be further explained by the loss of majority in committee (which correlates strongly with plenary support). Still, we introduce as a control a variable taking the value of 1 if there is a committee 19 For a more complete discussion of the advantages and dis-advantages of using cosponsorship data see Aleman et.al. (2009). 18

20 plurality and 0 otherwise. This variable distinguishes stacked committees in plurality-led congresses and committees of lesser importance that lack majority support in majority-led congresses. Legislator-specific control variables There are a number of variables that are of substantive interest and should be introduced to control for other confounding factors. Independent variables that facilitate the discharging of legislation to the floor include whether the sponsor is the president of the chamber or a committee chair. The number of years the sponsor has served in Congress (tenure) is also expected to improve success in committee. We also control for the natural log of the number of bills proposed by the sponsor in a congressional year. Because information is observed only at the level of the individual legislator, we add these in a second level. Bill-dependent control variables Other variables could affect important properties of the proposed bills. We control for the total number of legislators cosponsoring each bill; the number of committees that draft the majority report; whether legislators from more than one party support a bill (multi bloc); the type of legislative session (ordinaria, extraordinaria, prorroga); whether the bill was referred to one of three most important committees (Appropriations, Foreign Affairs, and Municipal Affairs); and dummy variables describing the party of the lead sponsor of the bill (PJ, UCR, FREPASO, etc). Notice that because we introduce party specific controls, our model is controlling for party specific effects. We expect that broader support will facilitate passage, as it is generally reflected by a larger number of cosponsors or when initiatives are supported by more than one bloc. In contrast, referral to multiple committees should increase the number of veto points and decrease success in 19

21 committee. Referral to important committees also should decrease success, because competition for more substantive legislation in higher ranked committees is more intense. The choice of committees also provides important information about the content of the bill. For example, any bill that includes spending measures needs to be reported by the Appropriations committee. Descriptive statistics reported in Table A.1 of the Appendix. In the next section we present four models of success in committee. In the first two models we estimate success in majority led congresses while distinguishing observations where the committee chair is a member of the majority party or a member of the opposition. We expect the loss of majority control to produce a mild ideological drift towards the median committee voter, consistent with the theory proposed in Section 2. The next two models estimates success in committees when majority control is lost, both for committees led by a chair from the plurality party or a minority chair. After estimating these four models of success in committee, we also report results from two models estimating success on the plenary floor, once the sample size has been restricted and includes only bills reported from committee. The model of legislative success in the floor allows us to analyze how changes in committee gatekeeping strategies affect overall legislative success. Legislative Success in Committee and on the plenary Floor Models A and C in Table 3 presents estimates of success in committees chaired by a member of the majority or plurality party. 20 Meanwhile, Models B and D describe the estimates of 20 In Table A.2 in the appendix we provide estimates of a restricted model including only the ideological distance variables. Alternative models to assess the robustness of estimates may be requested from the authors. 20

22 success in committees led by opposition chairs. Each of these models provides evidence that conforms to the theory proposed in section 2. << INSERT TABLE 3 >> Let us start by noticing the evolution of the estimates of ideological proximity to the median of the majority party and the median committee member across the different models. As is it possible to observe, ideological proximity to the median voter of the majority party is of the utmost importance for majority-chaired committees in majority-led congresses. As majority control is lost, however, the benefit to members of the majority party becomes less pronounced. Finally, in plurality led congresses with minority committee chairs (Model D), the proximity to the median voter of the majority party becomes statistically insignificant. By contrast, proximity to the median committee voter becomes a significant predictor of success. The effect of ideological proximity to the median committee voter of the majority party and the overall median committee member is more readily interpretable as presented in Figure 2. << INSERT FIGURE 2 >> The horizontal axis in Figure 2 describes the ideological location of legislators in the Argentine Congress as represented by its first dimension. The vertical axis describes rates of success in committee. Consistent with the model presented in the third section, the location of the median committee member of the majority-plurality party is represented with the letter M and the median committee voter is described with the letter C. As shown in the left plot of Figure 2, in committees chaired by a member of the majority party, success is higher for proposals sponsored by representatives that are deep in the majority party coalition. The loss of majority control results in a mild increase in the amount of legislation discharged by the median committee voter, as observed by the surface between the majority and plurality lines. 21

23 The right plot in Figure 2 describes success in committees led by an opposition chair. As can be observed, the total amount of legislation reported to the floor is larger than in committees with a majority party chair. The effect is particularly dramatic in committees chaired by the opposition in plurality-led congresses, where the median committee member is able to report 13% of bill initiatives e.g. 30% more than in majority led congresses. There are a number of other findings which deserve to be highlighted. First, it is worth noticing that multi bloc proposals result in an overall increase in success in committee. The positive and statistically significant effect of multi bloc proposals is also almost twice as large when majority control is lost. This finding holds when controlling for the party chairing the committee. The effect is not only statistically significant but also substantive: a bill cosponsored by members of different parties will have a 7% higher chance of being reported from committee in plurality-led congresses. Also noteworthy is that the positive effect of a larger number of cosponsors disappears in plurality-led congresses once we account for the effect of multi bloc proposals. Consistent with Krutz (2005) findings, we find that seniority has little effect on committee success. However, authority matters, as committee chairs are on average twice as successful in reporting bills from committee e.g. a success of 16% for chairs instead of the average 10% for other members. Referrals to more prestigious committees result in a decline in success, as shown by the negative coefficients for bills referred to the Appropriations and Municipal Affairs Committees. Such decline in success becomes more pronounced in plurality-led congresses, indicating that the change in the ideological make up of the legislation is also accompanied by a change in the makeup of the issues brought to the plenary floor. Bills initiated during regular congressional sessions (Sesiones Ordinarias) have on average a higher approval rate than those initiated in special sessions (Sesiones Prorroga or Extraordinarias). Such a finding is expected, given that special sessions are generally called to deal with pressing 22

24 bills, generally spearheaded by the Argentine President. Finally, all models in Table 3 show that bills initiated by parties other than the PJ or UCR have a lower chance of being discharged from committee. FREPASO sponsored bills will see their chances of success drop by 3% in both plurality and majority-led congresses. Similarly Provincial parties will see drops of 5% and 4% in plurality and majority congresses respectively. In Models E and F of Table 3 we present estimates of legislative success on the plenary floor, using the restricted sample of bills successfully reported from Committee. Results clearly show statistically significant change in the ideological make up of the legislation being approved in the Argentine House when majority control is lost. First, it is worth noticing that in majority-led congresses the ideological proximity to the median voter of the majority party on the plenary floor significantly increases overall legislative success. By contrast, when majority control is lost the ideological distance estimates become statistically insignificant, as legislation schedule by the Chamber Directorate now reflects further compromises among party leaders. The overall consequence is that in majority-led Congresses, the largest party controls both committee and floor outcomes, being able to restrict sequentially legislation that is not preferred by its median party voter at the committee level and in the Chamber Directorate. By contrast, the loss of majority control results in a change in the ideological make up of the legislation reported from Committee as well as in the legislation voted on the plenary floor. It is also noteworthy that legislation that was discharged from important committees has also a higher probability of receiving a favorable vote on the plenary floor. In other words, while legislation sent to important committees is less likely to be discharged, the smaller set that is successfully reported is indeed more likely to be approved. 23

25 Concluding Remarks In this article we provide extensive evidence that the loss of majority control leads to important changes in the scheduling strategies of majority and minority parties. We show that the loss of majority control significantly reduces the policy weight of the median committee voter of the plurality party and strengthens the position of the overall median committee member. Our findings open new insights into the strategies of committee chairs in majority- and plurality-led congresses. In contrast to Congressional research on the US House, we show that rational anticipation by Committee authorities is not an effective mechanism to restrict the set of legislation reported from committee in plurality-led congresses. In the particular case of the Argentine House, two other institutions, the pre-floor party meeting and the Chamber Directorate, are charged with the responsibility of preventing bills that divide the majority party from reaching the plenary floor. When majority control is lost, however, the Chamber Directorate is a critical institution where interparty agreements are negotiated. In this article, consequently, we provide a model that describes how contextual changes in the partisan make-up of the plenary result in different gatekeeping strategies. A further contribution of this article is to provide a first-of-its-kind analysis of legislator success in Committee in plurality-led congresses. While much research has sought to understand the determinants of legislative success in plenary votes, there is little research estimating the determinants of representatives committee success in the US and none that we are aware off for non-us Congresses. Our strategy to measure the gatekeeping authority through success in committee explains how parties control the flow of legislation sequentially, in pursuit of different goals in committee and on the plenary floor. Results presented in this article also provide evidence of a new and promising line of research, comparing models of legislative success subject to various partisan and contextual effects. 24